In this case, if you allow miners the liberty to include as large blocks as they like, an external actor that hates bitcoin can just say to a miner "you know what? I'll pay you more fees than what the others are making, and if your block is orphaned, I'll pay that too - just make sure to put in there as many txs as you like".
At that point, the enemy of bitcoin who has employed "bad miners" for very little money (compared to the multi-billion dollar interests of banks/payment companies/governments), can bloat bitcoin to death for extremely small costs.
Just because the miners (in this hypothetical) are allowed to create blocks as large as they desire, does not mean that other miners are forced to accept those blocks as valid. If such a large block (presumably the result of your bribe above) is deemed detrimental to the system by the other miners (i.e., detrimental to these other miners' interests), these other miners are incentivized to orphan that block. Resulting in no problem to the network. The 'problem' solves itself.
Not really.
1) The network is being spammed every day - even with stated spam attacks and stress tests. These blocks are now sitting in the blockchain as bloat. If the rejection scenario actually happened, they wouldn't be sitting in there.
Well, no. Any transaction that has made it into the blockchain is -- by definition -- a valid transaction. You don't get to call valid transactions 'bloat'.
Not to mention the fact that this has nothing to do with the scenario you posited to which I was replying.
2) Let's asssume I'm the exo-attacker and plan to do what I laid above. I also hire a programmer, modify bitcoin core to use GPU for validating big blocks, the ability to process blocks is raised significantly, and then release that software as open source so that miners will use it to "speed up new block verification and help scale bitcoin".
By employing GPU power *and* the fact that network propagation between the large miners is good, there is no valid reason for others to reject very big blocks because I've also given them the tools to process them. I mean the (other) miner is not going to need minutes to process them, so why not?
There is nothing wrong with your scenario 2. Other than the fact that you are completely changing the terms of the scenario re: the single attacker incentivized by an external payment that you opened with. If miners make bigger blocks, and other miners accept them as valid, then great. The system is able to accommodate more transaction volume. Wishful transactors don't need to shuffle off to alt coins in order to transactionate. The only possible problem here is if a critical number of validators can't keep up. Note that validators are free to employ the same techniques to improve their performance. Regardless, I say good riddance to a network hobbled by RasPis.
(this could happen too:)
3) The larger the miner, the biggest the incentive to accept even bigger blocks so that they can crush the opposition who can't handle them. It's like corporate cartel forming. You lower the price (to your detriment) until others fall out of the market. And then you raise the price. You can do the same with blocks. You accept the largest blocks possible, so that those who can't catch up fall behind. It's their problem, not yours. The less they mine, the more you earn.
Technological advancement rather akin to the CPU > GPU transition, the GPU > FPGA transition, the FPGA > ASIC transition, etc. That's worked out reasonably well in practice. Except this advantage is really not about size, but rather efficiency.
But really - you want to put some efficiency improvement numbers out for discussion? If it is more than a scant percent or two, would such not already be coded up and in the wild? Lacking evidence, reasoning suggests your 'crushing' speedup is not a potential significant possibility.
